Skip to content
AI Atlas
Changes

What changed in AI

Every material change the connectors observe becomes an event with a source. Dates are when the change occurred (effective date when known); hover a date for the observation time. Older events load with a cursor.

6,541 events since 11 Aug 2026 · ordered by occurrence

14 Sept 2026 1

  • AnnouncementCompanyOpenAI

    OpenAI: Perplexity trusts GPT-6 Astra with end-to-end systems

    openai

12 Sept 2026 49

  • Benchmark resultModelk2-horizon-375b-a23bMBZUAI Institute of Foundation Models

    k2-horizon-375b-a23b scores 1.52% on Terminal-Bench

    artificial_analysis
  • Benchmark resultModelk2-horizon-375b-a23bMBZUAI Institute of Foundation Models

    k2-horizon-375b-a23b scores 30.79 on Artificial Analysis Intelligence Index

    artificial_analysis
  • Benchmark resultModelZ.ai GLM 5.2Z.ai (Zhipu AI)

    Z.ai GLM 5.2 scores 1.01% on Terminal-Bench

    artificial_analysis
  • Benchmark resultModelZ.ai GLM 5.2Z.ai (Zhipu AI)

    Z.ai GLM 5.2 scores 34.01 on Artificial Analysis Intelligence Index

    artificial_analysis
  • Price changedModelLlama 3.1 70B InstructMeta AI

    OpenRouter changed pricing for Llama 3.1 70B Instruct: $0.72 in / $0.72 out per 1M tokens → $0.4 in / $0.4 out per 1M tokens

    $0.72 in / $0.72 out$0.40 in / $0.40 outopenrouter
  • Max output changedModelLlama 3.1 70B InstructMeta AI

    Llama 3.1 70B Instruct: max output tokens changed from 8192 to 16384

    Max output8.19K tokens16.4K tokensopenrouter
  • Price changedModelDeepSeek V4 Flash 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Flash 0423: $0.06594 in / $0.13188 out per 1M tokens → $0.06566 in / $0.13132 out per 1M tokens

    $0.066 in / $0.132 out$0.066 in / $0.131 outopenrouter
  • Price changedModelDeepSeek V4 Pro 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Pro 0423: $0.72819 in / $1.45638 out per 1M tokens → $1.6 in / $3.2 out per 1M tokens

    $0.728 in / $1.46 out$1.6 in / $3.2 outopenrouter
  • Price changedModelKimi K3Moonshot AI

    OpenRouter changed pricing for Kimi K3: $2.30273 in / $11.5502 out per 1M tokens → $2.64814 in / $13.2827 out per 1M tokens

    $2.3 in / $11.55 out$2.65 in / $13.28 outopenrouter
  • Max output changedModelDeepSeek V4 Pro 0423DeepSeek

    DeepSeek V4 Pro 0423: max output tokens changed from 384000 to 393216

    Max output384K tokens393.2K tokensopenrouter
  • Property changedFrameworkllama_indexLlamaIndex

    llama_index: kind changed from training-framework to agent-framework

    Kindtraining-frameworkagent-frameworkgithub
  • Property changedFrameworkllama.cppggml.ai

    llama.cpp: latest release at changed from 2026-09-12T10:59:16+00:00 to 2026-09-12T14:48:43+00:00

    Latest release12 Sept 202612 Sept 2026github
  • Property changedFrameworkllama.cppggml.ai

    llama.cpp: latest release tag changed from b10927 to b10931

    Latest release tagb10927b10931github
  • Version releasedFrameworkllama.cppggml.ai

    llama.cpp: latest version changed from b10927 to b10931

    Latest versionb10927b10931github
  • Property changedFrameworkpytorchPyTorch Foundation

    pytorch: latest release at changed from 2026-09-12T05:07:50+00:00 to 2026-09-12T06:38:55+00:00

    Latest release12 Sept 202612 Sept 2026github
  • Property changedFrameworkpytorchPyTorch Foundation

    pytorch: latest release tag changed from viable/strict/1789203000: [inductor] Make manual wait-use… to viable/strict/1789219412: Move c10/util/TypeCast.h and ov…

    Latest release tagviable/strict/1789203000: [inductor] Make manual wait-user repair linear (#195616)viable/strict/1789219412: Move c10/util/TypeCast.h and overflows.h into torch/headeronly (#194781)github
  • Version releasedFrameworkpytorchPyTorch Foundation

    pytorch: latest version changed from viable/strict/1789203000: [inductor] Make manual wait-use… to viable/strict/1789219412: Move c10/util/TypeCast.h and ov…

    Latest versionviable/strict/1789203000: [inductor] Make manual wait-user repair linear (#195616)viable/strict/1789219412: Move c10/util/TypeCast.h and overflows.h into torch/headeronly (#194781)github
  • New modelModelPhi-4-mini-instruct

    New model: Phi-4-mini-instruct

    microsoft_research
  • Price changedModelDeepSeek V4 Flash 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Flash 0423: $0.06622 in / $0.13244 out per 1M tokens → $0.06594 in / $0.13188 out per 1M tokens

    $0.066 in / $0.132 out$0.066 in / $0.132 outopenrouter
  • Price changedModelDeepSeek V4 Pro 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Pro 0423: $0.75864 in / $1.51728 out per 1M tokens → $0.72819 in / $1.45638 out per 1M tokens

    $0.759 in / $1.52 out$0.728 in / $1.46 outopenrouter
  • Price changedModelHy3Tencent

    OpenRouter changed pricing for Hy3: $0.132 in / $0.528 out per 1M tokens → $0.0825 in / $0.33 out per 1M tokens

    $0.132 in / $0.528 out$0.083 in / $0.33 outopenrouter
  • Price changedModelGLM 5.3 FlashZ.ai (Zhipu AI)

    OpenRouter changed pricing for GLM 5.3 Flash: $0.075 in / $0.25 out per 1M tokens → $0.15 in / $0.5 out per 1M tokens

    $0.075 in / $0.25 out$0.15 in / $0.50 outopenrouter
  • Openness changedModelStep 3.5 FlashStepFun

    Step 3.5 Flash: openness changed from proprietary to open-weights

    Opennessproprietaryopen-weightshuggingface
  • ReleaseFrameworkllama.cppggml.ai

    ggml-org/llama.cpp released b10931

    github
  • Price changedModelDeepSeek V4 Flash 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Flash 0423: $0.0665 in / $0.133 out per 1M tokens → $0.06622 in / $0.13244 out per 1M tokens

    $0.067 in / $0.133 out$0.066 in / $0.132 outopenrouter
  • Price changedModelDeepSeek V4 Pro 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Pro 0423: $0.768732 in / $1.53746 out per 1M tokens → $0.75864 in / $1.51728 out per 1M tokens

    $0.769 in / $1.54 out$0.759 in / $1.52 outopenrouter
  • Price changedModelDeepSeek-V4-Pro-0813DeepSeek

    OpenRouter changed pricing for DeepSeek-V4-Pro-0813: $0.57816 in / $1.73448 out per 1M tokens → $0.57948 in / $1.73844 out per 1M tokens

    $0.578 in / $1.73 out$0.579 in / $1.74 outopenrouter
  • Price changedModelDeepSeek V4 Flash 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Flash 0423: $0.06678 in / $0.13356 out per 1M tokens → $0.0665 in / $0.133 out per 1M tokens

    $0.067 in / $0.134 out$0.067 in / $0.133 outopenrouter
  • Price changedModelDeepSeek V4 Pro 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Pro 0423: $0.778824 in / $1.55765 out per 1M tokens → $0.768732 in / $1.53746 out per 1M tokens

    $0.779 in / $1.56 out$0.769 in / $1.54 outopenrouter
  • Property changedModelgrok-4-20-0309SpaceXAI

    grok-4-20-0309: reasoning changed from false to true

    ReasoningNoYesartificial_analysis
  • Property changedModelqwen3-omni-30b-a3b-instructAlibaba Group

    qwen3-omni-30b-a3b-instruct: reasoning changed from true to false

    ReasoningYesNoartificial_analysis
  • Property changedModelgrok-3SpaceXAI

    grok-3: reasoning changed from true to false

    ReasoningYesNoartificial_analysis
  • Property changedModelQwen3 Next 80B A3B InstructQwen

    Qwen3 Next 80B A3B Instruct: reasoning changed from true to false

    ReasoningYesNoartificial_analysis
  • Property changedModelQwen3 VL 8B InstructQwen

    Qwen3 VL 8B Instruct: reasoning changed from true to false

    ReasoningYesNoartificial_analysis
  • Property changedModelSonarPerplexity AI

    Sonar: reasoning changed from true to false

    ReasoningYesNoartificial_analysis
  • Property changedModelQwen3 MaxQwen

    Qwen3 Max: aa context window changed from 262144 to 256000

    Aa context window262,144256,000artificial_analysis
  • Property changedModelolmo-3-1-32b-instructAllen Institute for AI

    olmo-3-1-32b-instruct: aa context window changed from 65500 to 65536

    Aa context window65,50065,536artificial_analysis
  • Property changedModelStep 3.5 FlashStepFun

    Step 3.5 Flash: aa openness changed from proprietary to open-weights

    Aa opennessproprietaryopen-weightsartificial_analysis
  • ReleaseFrameworkllama.cppggml.ai

    ggml-org/llama.cpp released b10930

    github
  • Price changedModelDeepSeek V4 Pro 0423DeepSeek

    OpenRouter changed pricing for DeepSeek V4 Pro 0423: $0.799182 in / $1.59836 out per 1M tokens → $0.778824 in / $1.55765 out per 1M tokens

    $0.799 in / $1.6 out$0.779 in / $1.56 outopenrouter
  • Capabilities changedModelMiniMax-01MiniMax

    MiniMax-01: modalities changed from ["text"] to ["image", "text"]

    Modalitiestextimage, textopenrouter
  • Property changedModelMiniMax-01MiniMax

    MiniMax-01: modalities input changed from ["text"] to ["image", "text"]

    Input modalitiestextimage, textopenrouter
  • Capabilities changedModelQwen3.5-9BQwen

    Qwen3.5-9B: modalities changed from ["image", "text"] to ["image", "text", "video"]

    Modalitiesimage, textimage, text, videoopenrouter
  • Property changedModelQwen3.5-9BQwen

    Qwen3.5-9B: modalities input changed from ["image", "text"] to ["image", "text", "video"]

    Input modalitiesimage, textimage, text, videoopenrouter
  • Capabilities changedModelGemma 4 31BGoogle

    Gemma 4 31B: modalities changed from ["image", "text"] to ["image", "text", "video"]

    Modalitiesimage, textimage, text, videoopenrouter
  • Property changedModelGemma 4 31BGoogle

    Gemma 4 31B: modalities input changed from ["image", "text"] to ["image", "text", "video"]

    Input modalitiesimage, textimage, text, videoopenrouter
  • Capabilities changedModelGemma 4 26B A4BGoogle

    Gemma 4 26B A4B: modalities changed from ["image", "text"] to ["image", "text", "video"]

    Modalitiesimage, textimage, text, videoopenrouter
  • Property changedModelGemma 4 26B A4BGoogle

    Gemma 4 26B A4B: modalities input changed from ["image", "text"] to ["image", "text", "video"]

    Input modalitiesimage, textimage, text, videoopenrouter
  • Capabilities changedModelMiMo-V2.5Xiaomi

    MiMo-V2.5: modalities changed from ["text"] to ["audio", "image", "text", "video"]

    Modalitiestextaudio, image, text, videoopenrouter

Feed keyed on occurred_at. Backfill = history imported when a source is first crawled (excluded — tick the box in the rail to include it). Times are UTC.